Vol. 6, 993-997, December 1997 Cancer Epidemiology, Blomarkers & Prevention 993

Effect of Paternal and Maternal Cancer on Cancer in the Offspring:

A Population-based Study1

Karl Hemmlnki2 and Pauli Vaittinen

Department of Biosciences at Novum Karolinska institute, 14157 Huddinge

[K. H.], and Center for Epidemiology, National Board of Health and Welfare,

10630 Stockholm (P. V.], Sweden

Abstract

The Family-Cancer Database was constructed from thenationwide Swedish registries to include more than30,000 cancers in offspring diagnosed at ages 15-Si years

and their parents. Cancer risk in the offspring wasIncreased about 1.10 times when the father had cancer,whereas no increase was noted when the mother had

cancer. If both parents had cancer, the risk for sons was1.39 and for daughters, 1.34. FamIlial aggregation

between parents and offspring was observed for 5concordant and 14 discordant cancer sites and 10parental sites at which all cancer was increased in theoffspring. The concordant sites between the parent andoffspring were colorectum, breast, melanoma, skin(squamous cell carcinoma), and thyroid. The aggregationat discordant sites in the parents and the offspringincluded stomach-breast, colorectum-sailvary glands,

colorectum-breast, colorectum-lymphoma, colorectum-leukemia, liver-breast, pancreas-breast, breast-melanoma,

ovary-breast, prostate-breast, prostate-cervix, prostate-multiple myeloma, kidney-melanoma, and nervous tissue-melanoma. In most of these combinations, cancer in thesecond parent Increased the risk to the offspring. Thepresent results on young and middle-aged adults suggest

that cancer in both parents increases cancer risk in theoffspring at many sites. Chance and environmental effectsmay explain some of the results, whereas true genetic

factors probably contribute to most of the findings. Themolecular genetic explanation may be that rare dominantsingle genes increase susceptibifity at many sites or thatoverlapping sets of genes control susceptibility at multiplesfte�

Introduction

The study of familial clustering ofcancer has been fundamentalto the understanding of heritable components in cancer and thediscovery of the genes involved (1, 2). The concept of tumor

Received 3/1 1197; revised 7/7197; accepted 7/11/97.The costs of publication of this article were defrayed in part by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.I This study was supported by the King GustafV Jubileefund and by the SwedishCouncil for Planning and Coordination for Research.

2 To whom requests for reprints should be addressed, at Department of Bio-sciences, CNT Novum, 141 57 Huddinge, Sweden. Phone: 46-8-6089243; Fax:46-8-6081501; E-mail: knd.hemminki@cnt.ki.se.

suppressor genes that has evolved along with analysis of fa-milial cancers and family studies has been germane to themapping and characterization of most of the cancer suscepti-bility genes identified to date (2, 3). Although only 5% ofcancer is thought to be due to highly penetrant single-genemutations in the germ line, a much larger proportion of cancer

may involve somatic mutations in these genes in sporadic formsof cancer (2, 4). Additionally, it has become increasingly cvi-

dent that the hereditary cancer syndromes often entail an in-

crease in the risk of cancer at many sites other than the “index”sites, although at lower risk in the nonindex than in the index

site.Familial clustering of cancer has been studied most com-

monly after clinical identification of probands (5, 6). Thisapproach has been very productive in terms of understandingcancer genetics. Many forms of cancer in which a single geneposes a high risk have been identified. Some 200 single-genetraits are known in which cancer is a recognized complication(7). Another approach to the study of familial cancer has been

to analyze cancer risks of the relatives of the index case inanalytical epidemiological studies (8, 9). Twin studies offer a

third alternative for genetic epidemiology of cancer. Dissection

of heritable and environmental components is possible in suchstudies, and the risk estimates should be robust, but the rareness

of twinning impedes this approach (10-13). The fourth ap-proach to genetic epidemiology of cancer is a population-based

study in which all cancers are registered and family relation-

ships can be reconstructed. The power is in large numbers andunbiased risk estimates. These in turn allow estimation of

familiality at multiple sites. For gene-mapping purposes, thefamily units afford a possibility of applying allele-sharingmethods that require large population bases but are useful whenmany genes operate in the disease (14, 15). Population-based

studies have been carried out in a few geographic areas, in-cluding those on the Mormon population in Utah, which havebeen based on existing genealogy (16, 17). In Denmark and

Iceland, cancer cases have been obtained from the nationwidecancer registry, and family relationships have been constructed

from other national registers (18-20).Here, we present results from the population-based family

database from Sweden. The size of the population (8.7 million

in 1992) and the nationwide registration of cancer since 1958(1.4 million registered tumors) offer unique possibilities forepidemiological studies of cancer. The availability of a familydatabase on children born after 1940, including children and

their parents in the Second Generation Register, permitted

linkage to the Cancer Registry to form the Family-CancerDatabase. We believe that this database will be a useful re-

source for genetic epidemiology of cancer and identification ofnew underlying genes. We analyze here the risks of cancer inadult offspring of the parents with or without cancer. We wantto test the powers of the registered database in examiningcancer risks across multiple sites and in families in which both

on July 4, 2018. © 1997 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from

994 Family-Cancer Database

3 The abbreviations used are: RR, relative risk; CI, confidence interval.

Table 1 Cancer in offspring by parental cancer

Pare ntal cancer (father/mother) No. of children No. of cancers Adjusted RR 95% CI

Sons

Daughters

-I-

+1-

-1+

+1+

All

-I-

+1-

-1+

+1+

All

510,382121,771

99,918

26,606

758,677

490,31 1116,548

95,724

25,535

728.118

4,420

1,251

872

336

6,879

9,220

2,532

1,865

724

14,341

1.00

1.12”

0.97

1.39#{176}

1.00

1.07#{176}

0.98

1.34a

0.97-1.03

1.05-1.18

0.90-1.03

1.23-1.54

0.98-1.02

1.03-1.11

0.93-1.02

1.23-1.44

a Ninety-five % CI does not inc lude 1.00.

parents have cancer. We reason that modest inherited increases

in offspring cancer risk should be augmented when contributedby both parents.

Subjects and Methods

Registers and Source of Subjects. Statistics Sweden main-tans a Second Generation Register, in which the children bornin Sweden in 1941 and later are registered with their biologicalparents as families. By 1995, more than 6 million individuals ofthe population of 8.8 million were in the register. However, theregistration only took place for those children alive at theconstruction year of the register, 1992. In the l940s and 1950s,

each of the 5-year birth cohorts include more than 0.5 millionchildren. Considering children born between 1941 and 1955,97,600 of them were not included in the register because ofdeath. Some other reasons for not being included in the presentstudy were lack of one or both parents in the register (80,000children deleted) and some apparent technical mistakes. Thepopulation of children (here called offspring) included con-sisted of 486,650 (born 1941-45), 527,755 (1946-50), and472,390 (1951-55), a total of 1,486,795 children, together withtheir parents.

The Second Generation Register was linked by the indi-vidually unique national registration number to the Cancer

Registry. For the present study, only those children were in-cluded who were diagnosed with cancer at of 15 years or older.The highest possible diagnosis age, 51, in the offspring isreached by those born in 1941 if they were diagnosed after their

birthday during 1992; the common diagnosis ages for all of thethree 5-year birth cohorts were 17-36 years.

The nationwide Swedish Cancer Registry includes cancercases registered from 1958 and onward. Cancer registration is

considered to be close to 100% currently (21). Basal cellcarcinoma of the skin is not included in the registration. A

four-digit diagnostic code according to the 7th revision of theInternational Classification of Diseases is used. Cancers arealso recorded according to the first or subsequent primary

cancer and cancer in situ. The persons entered in the presentstudy were diagnosed for their first primary cancer during the

years 1958-1992 at ages 15-51 years. Cancer in situ was notincluded. Children diagnosed for their first primary cancerbefore the age of 15 years were excluded from the studypopulation.

Children born in 1941-1955 and alive at the end of 1992were divided into four cohorts according to the cancer status of

their both parents: -I-, neither parent had cancer; +1-, only

the father had cancer; -1+, only the mother had cancer; +1+both parents had cancer.

Analysis. The birth cohort-specific RRs� and the 95% CIswere calculated using the 5-year cohort-specific rate for off-spring in group -I- as the reference. The birth cohort-adjustedrates (“adjusted RR”) were calculated by the direct method.

Each of the three 5-year birth cohorts received an equal weight,according to the method used for the truncated European stand-

ard population (22). Indirectly, this method also makes anadjustment for age.

The 95% CIs were calculated supposing that the number ofcancer cases within a given time is Poisson distributed (22).

Results

A total of 21,220 cancers, diagnosed between ages 15 and 51,were recorded in the Family-Cancer Database among persons

born in 1941-1955. In the 5-year birth cohorts, 10,000 personsborn in 1941-1945 had cancer, as compared to 7,086 and 4,134persons in the subsequent 5-year cohorts. When the cancer riskwas analyzed by the parental cancer status, a systematic trend

was observed in all of the birth cohorts and in both sexes. If thefather had cancer but the mother did not (the +1- group), the

sons had a birth cohort-adjusted RR (adjusted RR) of 1.12 of

contracting cancer (Table 1). The offspring of two cancer-freeparents were the referents, with a RR of 1.00. Due to the largenumbers, this increase was statistically significant. When themother had cancer (the -1+ group), there was no excess risk inthe offspring. When both parents had cancer, the risk was 1.39,highly significant statistically. Somewhat lower RRs were ob-served for the daughters. However, the increases in the (+1-)and (+1+) groups of 1.07 and 1.34 were statistically signifi-cant.

The parental cancers in the Family-Cancer Database dis-tributed almost like all cancers in Sweden but with someskewing toward cancers detected at younger ages (21). Thethree most common paternal sites, based on the 7th revision ofthe International Classification of Diseases, were prostate, cob-rectum, and lung; common maternal sites were breast, coborec-

turn, and cervix uteri. Due to the age truncation, 15-51 years,the common cancers in the offspring were different. For sons,the Family-Cancer Database included 1103 testicular cancers,1063 melanomas, 858 lymphomas, 654 nervous system can-

on July 4, 2018. © 1997 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from

Cancer Epidemiology, Biomarkers & Prevention 995

Table 2 Cancer in off spring by pa ternal cancer site and by maternal cancer status

Cancer in father Cancer in offspring

Parental cancer status (father/mother)

Sons

+1- +1+

Daughters

+1- +1+ +1-

Both sexes

+1+

No. (RR) No. (RR) No. (RE) No. (RR) No. (RR) 95% CI No. (RR) 95% CI

Stomach Breast 59 ( 1.0) 25 ( 1 .9)” 59 ( 1.0) 0.8-1 .3 25 ( 1.9) 1.1-2.7”

Colorectum

Liver, bile

Salivary glands

Coborectum

Breast

Lymphomas

Leukemias

All sites

Breast

All sites

5 (4.8)

1 3 ( I .6)

17 (1.1)

2 (0.7)

150 (1.1)

26 (0.9)

0

9 (4.5)”

15 (4.4)”

1 (1.1)

58 (2.b)a

10 (1.4)

3 (2.6) 1 (3.7) 8 (3.6)

24 (2.0)” 4 ( 1 .4) 37 ( 1 .8)

1 10(1.1) 24 (1.1) 1 10 (1.1)

1 I (1.1) 1 (0.8) 28 (1.1)

2 (0.9) 1 (1.5) 4 (0.8)

314 (1.1) 86 (b.4)a 4�4 (1.1)

34 (1.5) 13 (2.6)” 34 (1.5)

73 (1.1) 25 (l.8)a 99 (1.1)

1.0-6.2” 1 (2.0)

1.2-2.4” 1 3 (2.7)

0.9-1.3 24 (1.1)

0.7-1.5 16 (3.1)

0.0-1.5 2 (1.2)10-12b 144 (1.6)

1.0-2.0” 13 (2.8)

0.8-1.3 35 (1.7)

1.2-4.3”

0.6-1.5j���7b

1.3-1.9”

1.1-4.5”11_23b

Pancreas Breast 47 (1 .7)” 9 (1 .5) 47 (1 .6) 1 . 1-2. 1” 9 (1 .5) 0.5-2.5

Lung All sites 153 (l.3)a 37 (1.3) 277 (1.1) 77 (1.3) 430 (1.1) 1.0-1.2” 1 14 (1.3) 1.0-1.5”

Prostate

Kidney

Breast

Cervix uteri

Multiple myelomaAll sites

Melanoma

I 1 (8.6)”

287(1.1)

6 (0.9)

082(l.5)a

2 (1.2)

227 ( I .2)” 73 ( 1 .5)” 227 ( 1 .2)

68 (1.0) 27 (1.9)” 68 (1.0)

2 (1.7) 1 (4.4) 13 (5.4)

6b4(b.l)a 180(1.3)” 901 (1.1)

21 (1.8)” 2 (0.7) 27 (1.5)

1 .0-1 .3” 73 ( 1 .5)

0.7-1.2 27 (1.9)2443b 1 (2.1)

10-12b 262(b.4)

0.9-2.0 4 (0.9)

1.2-1.9”

l.l�2.7”

1.2-1.5”

0.0-1.8

Melanoma Melanoma

All sites

9 (2.5)

25 (1.0)

3 (4.3)

9 (1.8)

10 (1.6) 5 (3.7) 19 (2.0)

54 (1.0) 17 (1.5) 79 (1.0)

1.1-2.9” 8 (4.0)

0.8-1.3 26 (1.6)

1.1-6.8”

1.0-2.3”

Skin Skin 8(6.5)” 0 2(2.1) 0 10(4.6) 1.7-7.5” 0

Nervous system Melanoma 8 (1.7) 1 (0.9) 15 (1.9) 5 (4.0) 23 (1.8) 1.1-2.6” 6 (2.6) 0.4-4.8

Thyroid gland Thyroid gland 2 (1 1.7) 0 1 (1.5) 1 (1 1.0) 3 (6.5) 1 (5.5)

Leukemias All sites 38 (1.1) 10 (1.3) 68 (1.0) 28 (1.7)” 106 (1.0) 0.8-1.2 38 (1.5) 1.0-2.1”

a Significant increase (95% CI does not include RR 1.00 of the -I- group).b Ninety-five % CI does not include 1.00.

cers, and 455 coborectal cancers. For daughters, breast cancerdominated, 4860 cases, followed by cervical cancer (1801

cases), melanoma (1694 cases), and ovarian cancer (919 cases);for lymphoma, there were 497 cases (data not shown).

The site-specific risk of cancer in the offspring (sons and

daughters separately and combined) was analyzed by 12 pater-nal and 15 maternal cancer sites. The cancer status of bothparents was considered in the +1-, -1+, and +1+ groups.This produced large amounts of data, but many of the cells in

this matrix were either empty or contained a few cases only.Some 37 statistically significant positive associations and 15negative associations were found. Tables 2 and 3 contain thosecombinations of sites at which a statistically significant positiveassociation was noted when the sons and the daughters wereconsidered separately or combined. All of the rates shown in

these tables are birth cohort-adjusted rates, and the (-I-)

group is taken as a referent.

Site-specific cancer in the offspring was analyzed by site-specific paternal cancer in Table 2. There was an increased riskof cancer in the offspring for colorectab and skin (squamouscell) cancer, concordant with the paternal site. Several increasesat discordant sites were observed, including father-offspringpairs for stomach-breast, coborectum-salivary glands, coborec-

tum-lymphoma, liver-breast, pancreas-breast, prostate-breast,

prostate-cervix uteri, prostate-multiple myeboma, kidney-mel-anoma, and nervous system-melanoma cancer sites.

An additional consideration in Table 2 was the modifica-tion of the cancer risk in the offspring by cancer status in thesecond parent (mother). In the +1- columns, mothers had no

cancer, whereas in the +1+ columns, mothers had various typesof cancer. There appeared to be a maternal enhancement of theeffect at many father-offspring cancer site combinations. En-hancement of cancer risk was noted, e.g., for father-son pairs in

coborectal-coborectal (RR of 1.6 in the +1- group as comparedto 4.5 in the +1+ group) and colorectal-lymphoma cancer sites(RR of 1.1 in the +1- group as compared to 4.4 in the +1+

group). A similar maternal effect was noted for father-daughter

pairs in stomach-breast, liver-breast, prostate-breast, and pros-tate-cervix cancer sites.

Site-specific cancer risk in the offspring was analyzed bysite-specific cancer in mothers (Table 3). Increased risk ofcancer was noted for the following concordant sites: breast,melanoma, and thyroid gland. A large number of case pairs forbreast cancer was noted: 229 and 90 for the breast-breastcomparisons in the -1+ and +1+ groups, respectively. In-creased risks at discordant sites (mother-offspring) were notedfor coborectum-breast, coborectum-beukemia, pancreas-breast,breast-melanoma, and ovary-breast. The risk of the offspringwas enhanced by paternal cancer in the mother-daughter pairs

of, e.g., coborectum-breast, pancreas-breast, breast-breast, and

ovary-breast sites.

Discussion

This is the first site-by-site analysis of the nationwide Swedish

Family-Cancer Database. The data are unique both in the sizeof the database and in its population-based structure. However,the Family-Cancer Database has two limitations because of theSecond Generation Register. One is that the data are from thoseborn in 1941 and later, causing truncation to persons ages 51

years and younger (Cancer Register was updated until 1992).Familial cancers are often recognized more clearly amongrelatively young adults, so the truncation does not invalidate theanalysis (23). The second limitation was that the Second Gen-eration Register lacked information from those born in 1941 orlater who had died before 1992. This caused a deficit in fatal

on July 4, 2018. © 1997 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from

996 Family-Cancer Database

Table 3 Cancer i n offs pring by ma temab cancer site and by patemal cancer status

Cancer in mother Cancer in offspring

Parental cancer status (father/mother)

+1-

Sons

+/+

Daughters

+/- +/+ +/-

Both sexes

+/+

No. (RR) No. (RR) No. (RR) No. (RR) No. (RR) 95% CI No. (RE) 95% Cl

Stomach Breast 21 (1.0) 6 (0.9) 21 (1.0) 0.5-1.4 6 (0.9) 0.2-1.7

Colorectum Salivary glands

Colorectum

Breast

Lymphomas

Leukemias

All sites

I ( 1 . 1 )

I I ( 1 .7)

19 ( 1.4)

2 (0.9)

1 18 (1.1)

0

6 (2.8)

6 (2.2)

3 (3.1)

38 (1.5)

3 (3.5) 1 (2.9) 4 (2.4)

9 ( 1 .0) 2 ( 1 . 1 ) 20 ( 1.3)

54 (0.8) 38 (l.9r 54 (0.8)

4 (0.6) 2 (1.4) 23 (1.1)

1 (0.5) 4 (6.2) 3 (0.7)

170 (0.8) 94 (l.5)a 288 (0.9)

0.0-4.9 1 ( I .6)

0.7-1 .9 8 ( 1 .8)

0.6-1.0 38 (1.9)

0.6-1.6 8 (1.9)

0.0-1.6 7 (4.4)

0.8-1.0 132 (1.5)

0.4-3.21225b

0.5-3.4

1.0-7.7”1218b

Liver, bile Breast

All sites 33 (1.0) 16 (1.7)

22 (0.9) 8 (1.1) 22 (0.9)

60 (0.8) 26 (1.2) 93 (0.9)

0.5-1.2 8 (1.1)

0.7-1.1 42 (1.3)

0.3-1.8

0.9-1.8

Pancreas Breast 19 (1.0) 14 (2.3)a 19 (1.0) 0.5-1.5 14 (2.3) 10�35b

Lung All sites 35 (1.0) 12 (1.3) 70(0.9) 33 (l.6)a 105 (1.0) 0.8-1.2 45 (1.5) 10�19b

Breast Breast

Melanoma

All sites

42 (1.3)

223 (1.0)

15 (1.5)

78 (1.3)”

229 (1.5)” 90 (2.lr 330 (1.5)

51 (0.9) 22 (1.5) 93 (1.0)

5l� (1.1) 192 (l.5)a 729 (1.1)

1.3-1.6” 91 (2.1)

0.8-1.3 37 (1.5)10-11b 270 (1.4)

1�25b

1.0-2.0”

1.2-1.6”

Cervix uteri All sites 38 (0.7) 18 (1.4) 1 18 (1.0) 43 (1.3) 156 (0.9) 0.8-1.1 61 (1.4) 10�1,7b

Corpus uteri All sites 68 (1.2) 14 (1.0) 1 1 1 (0.9) 46 (l.5)a 179 (1.0) 0.9-1.2 60 (1.3) 1.0-1.7”

Ovary

Kidney

Breast

All sites

Melanoma

47 (0.9)

5 (1.1)

22 (1.6)

0

42 (1.1) 22 (1.9)” 42 (1.1)

129 (1.1) 42 (1.4) 176 (1.1)

6 (0.9) 2 (0.8) 1 1 (1.0)

0.8-1.4 22 (2.0)

0.9-1.2 64(1.4)

0.4-1.6 2 (0.5)

1.1-2.8”

10�18b

Melanoma Melanoma

All sites

I 1 (3.5Y’

30 (1.4)

2 (2.0)

6 (1.1)

13 (2.4)� 5 (3.6) 24 (2.8)

57 (1.3) 14 (1.2) 87 (1.3)

1.7-3.9” 7 (3.1)

10�16b 20 (1.1)

0.7-5.4

0.6-1.6

Skin Skin 0 0 0 0 0 0

Nervous system Melanoma 4 (0.8) 0 10 (1.2) 4 (2.2) 14 (1.1) 0.5-1.7 4 (1.4) 0.0-2.8

Thyroid gland Thyroid gland 5 (1 1.8) 1 (10.1) 6 (4.0) 3 (7.8) 1 1 (5.8) 2.3-9.3” 4 (8.3) 0.0-16.6

Leukemias All sites 23 (1.1) 8 (1.2) 44 (1.0) 17 (1.4) 67 (1.0) 0.8-1.3 25 (1.3) 0.8-1.9

“ Significant increase (95% Cl does not include RR = 1.00 of the -/- group).b Ninety-five % CI does not include I .00.

cancers, particularly those of lung, pancreas, and liver, andfamiliality in these cancers may not be recognized. However,

this is not a cause of spurious positive associations. Anotherinherent limitation of the present kind of data is the inability todistinguish genetic and environmental effects. Families share

many environmental causes of cancer, including diet, smokinghabits, and many other lifestyle factors, warranting caution in

the interpretation of the results.Many cancer syndromes were initially recognized based

on large excess risk at particular sites (23). However, extendedstudies often revealed increased risks at sites other than the“index” site, e.g., Li-Fraumeni syndrome, and early-onset

breast cancer (2, 4, 7, 23, 24). The molecular explanation forthis has been the operation of the same susceptibility (tumorsuppressor) gene, such as Rb, p53, BRCAJ, and BRCA2, in

several types of cancers (2, 5). Interestingly, large, population-based studies have revealed clear familial risks between dis-cordant cancer sites, such as breast, colon, and prostate; nerv-ous tissue and melanoma; and breast and thyroid in the UtahPopulation Database (17). Similarly, in the largest cancer study

published on twins, there was an approximately 1 .5-fold excessrisk of all cancer in monozygotic as compared to dizygotic

twins (13).The cancer sites showing a statistically significant increase

in the offspring if one or both parents had cancer (based onTables 2 and 3) are listed in Table 4. The assumption of a

familial component would be strengthened if the increased RRswere seen in both sexes and in several parental groups. Familial

aggregation was evident in 19 pairs of cancer sites and in 10combinations in which all cancer in the offspring was in-

creased. Some combinations only appeared as solitary findings

and may be spurious. Because many comparisons were done toproduce Table 4, chance associations cannot be excluded. How-

ever, given that we observed a total of 37 positive associations

and 15 negative associations, it appears prudent to assume that

a large proportion of the associations has a biological basis.

The concordant sites in the parents and in the offspringwere colorectum, breast, melanoma, skin (squamous cell car-

cinoma), and thyroid gland (Table 4). Familiality is known at

most of these sites, and some of the susceptibility genes have

been discovered, including adenomatous polyposis coli and

mismatch repair genes in colon cancer, BRCAJ and BRCA2 inbreast cancer, pitS in melanoma (25) and ret in thyroid cancer

(26). In squamous cell carcinoma of the skin, p53 is ofteninvolved, but other genes have yet to be identified.

The discordant sites, shown in Table 4, included coborec-

turn-breast, ovary-breast, and prostate-breast, detected in 5ev-

eral previous studies (16, 18, 23). Other combinations, cob-

rectum-leukemia and nervous tissue-melanoma, showed

familial aggregation also in the Utah study (16). The remaining

discordant combinations, stomach-breast, coborectum-salivary

glands, coborectum-lymphoma, liver-breast, pancreas-breast,

breast-melanoma, prostate-cervix, prostate-multiple myeloma,

and kidney-melanoma, may be novel.

This study was designed to examine the effect of paternal

and maternal cancer individually and in combination on the

cancer risk of the offspring. The combined parental effect on

cancer risk in the offspring has seldom been a subject of study.

One of the likely reasons is that in site-by-site analysis, most

commonly exercised, the number of subjects is rarely large

enough; another reason may be that many common cancers are

on July 4, 2018. © 1997 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from

Colorectum

Breast

Melanoma

Skin

Thyroid gland

Breast

Salivary glands

Breast

Lymphoma

Leukemia

Breast

Breast

Melanoma

Breast

Breast

Cervix

Multiple myeloma

Melanoma

Melanoma

All cancer

All cancer

All cancer

All cancer

All cancer

All cancer

All cancer

All cancer

All cancer

All cancer

Present

Present

Weak

No cases

No

Present

Few cases

Present

Present

Present

Present

Weak

Present

Present

Present

Present

Few cases

No

Weak

Present

Present

Weak

Present

Present

Present

Present

Present

Weak

Present

Cancer Epidemiology, Blomarkers & Prevention 997

Table 4 Cancer sites where a familial effect or reinforcement of the effect bycancer in the second parent are noted, compiled from Tables 3 and 4

Cancer in parent Cancer in offspring Reinforcement

Same site

Colorectum

Breast

Melanoma

Skin

Thyroid gland

Different site

Stomach

Colorectum

Colorectum

Colorectum

Colorectum

Liver, bile

Pancreas

Breast

Ovary

Prostate

Prostate

Prostate

Kidney

Nervous tissue

All cancer in offspring

Colorectum

Liver, bile

Lung

Breast

Cervix uteri

Corpus uteri

Ovary

Prostate

Melanoma

Leukemia

sex specific, and analysis of a combined parental effect at a

particular site is not relevant.An interesting observation in this study was the increase in

cancer risk in the offspring when both parents had cancer. TheRR of all cancer was 1.39 for sons and 1.34 for daughters. The

enhancement of risk in the offspring by cancer in both parentswas observed for most of the sites listed in Table 4. The effect

was observed for the parent-offspring combinations at the con-cordant sites, such as coborectum-coborectum and breast-breast,

and at discordant sites.The present data are in accordance with reports that there

is a mechanistic link between sets of cancer sites, probablybecause the same gene/genes are involved. The tumor suppres-

sor genes (and oncogenes) known to date act in a dominantfashion (1 , 4). Assuming the operation of a single gene con-trolling cancer risk at a few sites, the results would be com-patible with rare dominant alleles being inherited from both

parents and thus causing an increase in a few sites in theoffspring. The increase in risk would be rather small in thismodel and is unlikely to explain most of the findings in the

present study. As another alternative, many partially overlap-ping dominant genes may control cancer at multiple sites. Thus,inheriting mutant alleles to any of this set of genes would causean increase of cancer risk at multiple sites as observed. Further

refinement of the data in the Family-Cancer Database shouldallow formal testing of such alternatives. The largeness of the

database should also make segregation analysis and comparison

of the models of inheritance possible for a number of cancer

sites.

References

I . Knudson, A. G., Jr. Hereditary cancer, oncogenes, and antioncogenes. Cancer

Res., 45: 1437-1443, 1985.

2. Li, F. P. Phenotypes, genotypes, and interventions for hereditary cancers.Cancer Epidemiol. Biomarkers Prey., 4: 579-582, 1995.

3. Collins, F. Positional cloning moves from perditonal to traditional. Nat.

Genet., 9: 347-350, 1995.

4. Li, F. P. Familial cancer syndromes and clusters. Curr. Probl. Cancer, 14:

75-106, 1990.

5. Easton, D. F. The inherited component ofcancer. Br. Med. Bull., 50: 527-535,

1994.

6. Lynch, H. T., Fusaro, R. M., and Lynch, J. Hereditary cancer in adults. CancerDetect. Prey., 19: 219-233, 1995.

7. Connor, J. J., and Ferguson-Smith, M. A. Essential Medical Genetics. Oxford,

UK: Blackwell Publishers, 1993.

8. Carter, B. S., Beaty, T. H., Steinberg, G. D., Childs, B., and Walsh, P. C.

Mendelian inheritance offamilial prostate cancer. Proc. NatI. Acad. Sci. USA, 89:

3367-3371, 1992.

9. Isaacs, S. D., Kiemeney, L. A. L. M., Baffoe-Bonnie, A., Beaty, T. H., andWalsh, P. C. Risk of cancer in relatives of prostate cancer probands. J. NatI.Cancer Inst., 87: 991-996, 1995.

10. Hrubec, A., and Ned, J. V. Contribution of familial factors to the occurrenceof cancer before old age in twin veterans. Am. J. Hum. Genet., 34: 658-671,

1982.

1 1. Braun, M. M., Caporaso, N. E., Page, W. F., and Hoover, R. H. Genetic

component of lung cancer: cohort study of twins. Lancet, 344: 440, 1994.

12. Braun, M. M., Caporaso, N. E., Page, W. F., and Hoover, R. H. A cohort

study of twins and cancer. Cancer Epidemiol. Biomarkers Prey., 4: 469-473,1995.

13. Ahlbom, A., Lichtenstein, P., Malmstrdm, H., Feychting, M., Hemminki, K.,

and Pedersen, N. Genetic and familial risk factors for cancer in twins. J. Natl.Cancer Inst., 89: 287-293, 1997.

14. Khoury, M. J., Beaty, T. H., and Cohen, B. H. Fundamental of GeneticEpidemiology. Oxford. UK: Oxford University Press, 1993.

15. Lander, E. S., and Schork, N. J. Genetic dissection ofcomplex traits. Science

(Washington DC), 265: 2037-2048, 1994.

16. Cannon-Albright, L. A., Thomas, A., Goldgar, D. E., Khosrow, G., Rowe, K.,Jacobsen, M., McWhorter, W. P., and Skolnick, M. H. Familiality of cancer inUtah. Cancer Res., 54: 2378-2385, 1994.

17. Goldgar, D. E., Easton, D. F., Cannon-Albright. L. A., and Skolnick, M. H.

Systematic population based assessment of cancer risk in first degree relatives ofcancer probands. J. NatI. Cancer Inst., 86: 1600-1607, 1994.

18. Tulinius, H., Egilsson, V., Olafsdottir, G. H., and Sigvaldason, H. Risk of

prostate, ovarian and endometrial cancer among relatives of women with breast

cancer. Br. J. Med., 305: 855-857, 1992.

19. Olsen, J. H., Boice, J. D., Seersholm, N., Bautz, A., and Fraumeni, J. F., Jr.Cancer in the parents of children with cancer. N Engi. J. Med., 333: 1594-1599,

1995.

20. Carstensen, B., Soll-Johanning, H., Villadsen, E., Sodergaard, J. 0., and

Lynge. E. Familial aggregation of colorectal cancer in the general population. Int.J. Cancer, 68: 428-435, 1996.

21 . Center for Epidemiology. Cancer Incidence in Sweden. Stockhom: NationalBoard of Health and Welfare, 1995.

22. Esteve, J., Benhamou, E., and Raymond, L. Statistical Methods in CancerResearch. IV. IARC Scientific Publ. No. 128. Lyon. France: IARC. 1994.

23. Tomatis, L. Cancer: Causes, Occurrence and Control. IARC Scientific PubI.No. 100. Lyon, France: IARC, 1990.

24. Li, F. P., Fraumeni, J. F., Jr., Mulvihill, J. J., Blattner, W. A., Dreyfus, M. G.,

Tucker, M. A., and Miller, R. W. A. A cancer family syndrome in twenty-fourkindreds. Cancer Res., 48: 5358-5362, 1988.

25. Cannon-Albright, L. A., Kamb, A., and Skolnick, M. A review of inheritedpredisposition to melanoma. Semin. Oncol., 23: 667-672, 1996.

26. Eng, C., Mulligan. L. M., Healey, C. S., Houghton, C., Frilling, A., Raue, F.,

Thomas, G. A., and Ponder, B. A. J. Heterogeneous mutation of the RE!’

proto-oncogene in subpopulations of medullary thyroid carcinoma. Cancer Res.,56: 2167-2170, 1996.

on July 4, 2018. © 1997 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from

1997;6:993-997. Cancer Epidemiol Biomarkers Prev   K Hemminki and P Vaittinen  offspring: a population-based study.Effect of paternal and maternal cancer on cancer in the

  Updated version

  http://cebp.aacrjournals.org/content/6/12/993

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cebp.aacrjournals.org/content/6/12/993To request permission to re-use all or part of this article, use this link

on July 4, 2018. © 1997 American Association for Cancer Research. cebp.aacrjournals.org Downloaded from